Hello. Working on a piece of genetic programming, and I have the situation where I want to have a set of symbols that will be placed in a list. Later when I am evalling that list as a function, I would like to be able to set the values of those symbols, with something like a let

example my program evolves a function (* (- 4 x) y). I would like to be able to provide values for x and y when I am running my fitness function

that is a function

(an expression parameterized over some values)

Yes.

Except I don't have values for x and y at first, and need to provide/define them later

but that is a function

(fn [x] x) is "I don't have a value for x and will provide it later"

Ah, I think I see.

Okay, no, that is not it. I am trying to evolve programs, so I have a function that builds nested lists, ensuring that the first argument is a function, and any subsequent list items are either valid s-expression or terminals like 1

thus the whole thing can be evaled and produce a result

I'm seeking a way to do two things. One, include terminals like x or y in the set of possible chosen values. Two, be able to call an evolved function, with the values of x and y present.

I would like to avoid using state, and having every 'x' terminal be "look up the value in an atom" but that could work

it seems refs or vars are close, but then I am unsure of how to use them precisely. I'm going to go read up on them

those are the same thing

not refs and vars, but the two things you want to do

... guess that's why i need to read up on them. 😓

since you are building up a clojure form to be evaled, what stops you from wrapping the whole list in a (let [x ... y ...] ...)

or a defn and then calling the function

you want to be able run some code multiple times with different values for x and y

in some cases changing the values, in other cases not

make your code a function and pass in x and y

I'll try and hammer on it for a while, see if the lightbulb pops.

Thank you both though, I hope I do not sound ungrateful.

you could use with-local-vars too (which is almost never used in Clojure)

What is the reason for not being able to use java static functions like clojure functions? As an example, I want to do this: (every? Character/isUpperCase [\a \b \c]), but am told that Clojure is “Unable to find static field isUpperCase in class java.lang.Character”. consequently I have to wrap it like this: (every? #(Character/isUpperCase %) [\a \b \c])

An unsatisfying but true answer is "Because Rich Hickey designed it that way", leading immediately to the question "Why did Rich Hickey design it that way?" I do not know the answer to that question, but suspect it may have something to do with the JVM mechanisms chosen for function calling, versus straight JVM method calls, and efficiency.

Or perhaps also the flexibility rather than efficiency of Clojure function calls, i.e. having variable number of args.

A little unsatisfying 😄 but thanks

https://stackoverflow.com/a/35200594/4379329 has a decent explanation

Ah that’s interesting and explains a lot. Thank you!

with MethodHandles now, it might be more feasible to build this into the compiler directly now

I have an issue with a quiz from 4clojure http://www.4clojure.com/problem/55 My solution to the problem works on my host. user=> (defn solved? [f] (= (f [1 1 2 3 2 1 1]) {1 4, 2 2, 3 1})) #'user/solved? user=> (solved? #(->> % (group-by identity) (into {} (map (fn [[k v]] [k (count v)]))))) true However, the function #(->> % (group-by identity) (into {} (map (fn [[k v]] [k (count v)])))) doesn't pass on 4clojure website with the following error message: clojure.lang.ArityException: Wrong number of args (1) passed to: core$map Is it a version issue? Any idea?

Very likely http://4clojure.com is running an older version of Clojure that doesn't have transducers.

Based on https://github.com/4clojure/4clojure/blob/develop/project.clj it seems to be running Clojure 1.4 perhaps? I would have expected someone to have updated it a bit in the last six years... No sure who maintains it these days.

Thanks @seancorfield Could you suggest me another implementation for the problem without using transducer?

Sure, just transpose the into/`map` calls to avoid that arity.

(solved? #(->> xs (group-by identity) (map (fn [[k v]] [k (count v)])) (into {})))

(not sure if I typed that right, but that's the general idea)

user=> (solved? #(->> % (group-by identity) (map (fn [[k v]] [k (count v)])) (into {}))) true

I will dig more into what the transducers are. Thank you very much 🙂

Your (into {} (map (fn ...))) form gets the expression threaded into the last argument position of the into call, but your map call only has one arg -- the function -- so that is valid with transducers (but was invalid before).

So you get (into {} xform expression) where xform is the transducer returned by the 1-arg map call.

Perfectly explained. I didn't even mean to use transducers when I came up with my previous implementation. Now I understand that I accidentally used the form of (into {} xform expression). Although I met your version of implementation without transducers to workaround the old version of 4clojure, I think yours look more elegant and simpler. Thanks for your lesson.

BTW, can I run a clojure with a specific version with lein repl, say, clojure 1.4.0 on my host for a testing and debugging purpose?

hey guys, i want to play around with clojure spec, so i've generated a project with lein, my problem is that i can't get the repl to invoke functions i've created 😞

(ns demo-spec.core
  (:gen-class)
  (:require [clojure.spec :as s]
            [clojure.spec.test :as test]))

(defn my-index-of
  "Return the at which search appear in source."
  [source search]
  (str/index-of source search))

(my-index-of "foobar" "b")
(apply my-index-of ["foobar" "b"])

(defn -main
  "I don't do a whole lot ... yet."
  [& args]
  (println "Hello, World!"))

(defproject demo-spec "0.1.0-SNAPSHOT"
  :description "FIXME: write description"
  :url ""
  :license {:name "Eclipse Public License"
            :url ""}
  :dependencies [[org.clojure/clojure "1.10.0"]]
  :main ^:skip-aot demo-spec.core
  :target-path "target/%s"
  :profiles {:uberjar {:aot :all}})

in project root i ran lein repl but can't call any of the functions 😞

You did not see any error messages when you ran lein repl? When I try to reproduce what you have done, it says there is no such namespace clojure.spec, which should be instead clojure.spec.alpha, and similarly clojure.spec.test.alpha. Also there should be a require of [clojure.string :as str] or else the str/index-of symbol is not resolved.

no

[clojure.main main main.java 37]]}
nREPL server started on port 46065 on host 127.0.0.1 - 
REPL-y 0.3.7, nREPL 0.2.12
Clojure 1.10.0
Java HotSpot(TM) 64-Bit Server VM 1.8.0_191-b12
    Docs: (doc function-name-here)
          (find-doc "part-of-name-here")
  Source: (source function-name-here)
 Javadoc: (javadoc java-object-or-class-here)
    Exit: Control+D or (exit) or (quit)
 Results: Stored in vars *1, *2, *3, an exception in *e

What is above the first line you pasted?

That looks like the last line of a stack trace, which I saw, where the first few lines had an error message

humm

If you see lines like that, it can be a sign of a problem.

thank you 🙂

i'll make some changes and see

sure, np

For reference: I ended up with a solution to my issue like so

(defn expression-to-function [arguments expression]
  (eval (list 'fn arguments expression)))

Also this isn't necessarily a problem, but you should be able to do experimentation without using (:gen-class), aot, or mentions of uberjar in your project.clj file. You might need such things if you want to deploy a compiled Clojure application, but for REPL experimentation they are not necessary, and can in some scenarios cause issues.

ok i got it

@andy.fingerhut i did those changes you mentioned and it's working fine now, thank you 🙂

cool

Is there any handy way to transform a map with namespaced keys, like this:

#:funky-cool-aid {:color :purple}

{:color :purple} 

(reduce-kv (fn [m k v] (assoc m (keyword (name k)) v)) {} your-map) should do it

Or (into {} (map (juxt (comp keyword name) identity) your-map)) if you're feeling sassy 🙂

@seancorfield thank you, I'll have a look at both, it will be used quite frequently so i'll add them to danger-util ✌️

(I should resist trying to write a point-free version at ten at night)

7:04 in sweden 😅

am

My second attempt was wrong -- I'd need a bit more in there

(into {} (map (juxt (comp keyword name key) val) your-map)) maybe something like that... I should open a REPL back up...

i just tried it, it works with the updated version 🙂

user=> (def your-map  #:funky-cool-aid {:color :purple :taste :grape})
#'user/your-map
user=> your-map
#:funky-cool-aid{:color :purple, :taste :grape}
user=> (into {} (map (juxt (comp keyword name key) val) your-map))
{:color :purple, :taste :grape}
user=>   

Although I think it's more that Grape Kool-Aid tastes "purple flavor" 🙂

I don't advise the point-free style really. Just wanted to show it was possible.

👍 thank you for the time @seancorfield

I have a map where each value is an array of ints. For example: {:a [1 2 3], :b [4 5 6]} I want to update this map so that the arrays no longer contain even numbers: {:a [1 3], :b [5]}. What is the idiomatic way of doing this?

@r.natarajan35 one way of processing maps keys and values is to use reduce-kv e.g

(reduce-kv (fn [m k v] (assoc m k (filterv even? v))) {} {:a [1 2 3], :b [4 5 6]})

oh wait, no longer contain even numbers

s/even?/odd? in the above

another common approach is to use map-vals, although it's not in core, it's commonly implemented

or a list comprehension, see this stackoverflow answer https://stackoverflow.com/questions/1676891/mapping-a-function-on-the-values-of-a-map-in-clojure

@danieleneal Oh wow, thanks! I really like the list comprehension version, it makes what i'm doing very clear to read

what is the recommended linter and formatter tool ?

@r.natarajan35 just as an alternative, here’s a Specter solution:

user=> (def a {:a [1 2 3], :b [4 5 6]})
#'user/a

user=>
user=> (setval [MAP-VALS ALL even?] NONE a)
{:a [1 3], :b [5]}

Question about destructuring of nested maps - why the outermost key is specified last in such expressions? E.g., let's say I have

(def data {:platform "Windows" :type {:name "2008 R2" :edition "Standard"}})

(let [{{:keys [name edition] } :type} data]
(println name edition ))
2008 R2 Standard

(let [{:type {:keys [name edition] }} data]
(println name edition ))
Syntax error macroexpanding clojure.core/let at (form-init5616907436889296336.clj:1:1).
[...]

@sy_borg that’s the syntax described here: https://clojure.org/guides/destructuring#_associative_destructuring

I've come just from there

(let [[{{:keys [name edition] } :type} data] ..): it takes the value of :type in data and then you do an additional destructuring on that

I get it, I was confused about the order

Yeah, map destructuring is done in pairs, and in each pair the thing you're destructuring is on the right hand side and the symbols/names that you want to bind it to are on the left hand side. This is similar to how in a let binding, the thing is on the right hand side, and the name on the left, you'd say (let [a 3] a)

@danieleneal yes, I would think so if it was about data symbol position, but it's about a key

Hey, there!

(ns my-app.database)

(def database (atom {}))

(defn reset-database [] (reset! database {}))

(defn add-record [table record]
  (swap! database update-in [table] conj (into {} record))
  {:ok record})

(defn remove-record [table record]
  (swap! database update-in [table]
         (fn [records]
           (remove #(= record %) records))))

(defn get-all-records [table] (get @database table))

(defn get-record-by-id [table id]
  (->>  (get-all-records table)
        (filter #(= id (:id %)))
        first))

(defn update-record [table id update]
  (let [record (get-record-by-id table id)
        updated-record (merge record update)]
    (swap! database update-in [table] record updated-record)))

I just wrote this atom to have a simple in-memory database

but if I use a record in the update-record

it says that it can't use a record as a Lang.FN

what I'm doing wrong?

swap turns into basically (update-in @database [table] record updated-record)

update-in expects a fn as 3rd arg

you might mean to put record in that vector with table

I have a vector with maps

and I want to update a value inside this map

but apparently doing this is not usual in clojure

and you probably want assoc-in

I have ({:id 1 :name "Gabriel", {:id 2 :name "John"})

and want to change name to "Jane"

where id is 2

but I could not figure it out

without using a map

change your db to use the id as an index and it would be alot easier but yea i think you would need to use a map

you mean using {id {record}}

yes. provided id is a unique value

try using {1 {:name "Gabriel"} 2 {:name "John"}}

sure

but the other way works too, you’d just need to use a (map ...) that checks if the id is correct then run the update when it is correct.

I would advise against using map in that case. Map will iterate over every single entry in the DB.

yea

If I’m writing a small utility to replace all :smile: shortcodes in a markdown document with the emoji characters, how should I be parsing it? Is there a parser combinator library someone could recommend or should I use a line-seq and regex?

Check out Instaparse (and the #instaparse channel)

https://github.com/Engelberg/instaparse

Hello, I am having issues running a Leiningen project in the repl within vscode via the Clojure extension. Where should I start to troubleshoot this? version: Leiningen 2.8.3 on Java 1.8.0_191 OpenJDK 64-Bit Server VM error:

Warning: implicit middleware found: cider-nrepl.plugin/middlewarePlease declare all middleware in :middleware as implicit loading is deprecated.

Is there a good analog to mocha's only method (https://mochajs.org/) in cljs.test (or clojure.test for that matter)? I have ~100 tests and I'm getting tired of commenting out un-used tests and/or changing the regular expression on my run-all-tests call.

While I'm at it, is there a good alternative to cljs.test? I'm pretty unhappy with it in a number of aspects, and I'm on the cusp of writing my own testing framework 😕

You could check out kaocha

oh yess thank you

that looks so much better

Looks like clojurescript support is pretty new, any idea if it works with shadow-cljs?

Or at least, code written for shadow-cljs

@jstaab what are you unhappy about?

if you want to run a single test, you could give https://github.com/Olical/cljs-test-runner a try

say i really wanted to check whether a value was false like so:

(s/explain #{false true} false)
val: false fails predicate: :clojure.spec.alpha/unknown
:clojure.spec.alpha/spec  #{true false}
:clojure.spec.alpha/value  false

@mfm false?

the fact that sets work as predicates relies on the set returning the arg if present, this doesn't work for nil or false

ah contains?

you can use #(contains? #{false true} %} but that doesn't give you a free generator like a set does

#(contains? #{true false} %)yeah

@borkdude the spec should fail if it's any value other than false and true

yeah thanks

you can use boolean? here of course

(s/with-gen ... (fn [] boolean?))

Why not boolean? Just a contrived example for learning?

maybe boolean? already has a generator defined? my spec knowledge is a little rusty

i'd like to use this as output for less-technical people, so i want really explicit error messages

I think "boolean?" is pretty self explanatory

i tried expound, but it didn't explicate the acceptable values for boolean?

(not that it should have to)

i think it is self-explanatory, but i'm trying to be very careful

@borkdude Some of the trouble is incidental and because of shadow-cljs, but off the top of my head: - No way to target individual tests without writing code. You can target namespaces with a regular expression or run-tests, but I'd really love command-line or local-code selection mechanisms (like kaocha offers) - I've had trouble getting tests to exit with a 1 for ci purposes. I can't remember how I got that working, but I may have wrapped them with a bash script that parsed the output - There aren't a lot of nice assertion methods. Not a big deal; it's easy to write your own, but is= is the big omission here. - Unless this was an artifact of shadow-cljs's test runner, stack traces get swallowed. If there's an error, I want to see the whole thing. As a result, I'm passing in a custom formatter that calls *print-err-fn*.

@mfm Hum, there's a library for that, let me try an find it. Because shoe-horning simpler spec seems wrong.

interesting! thanks

@mfm

ClojureScript 1.10.439
cljs.user=> (require '[clojure.spec.alpha :as s])
nil
cljs.user=> (s/explain boolean? true)
Success!
nil
cljs.user=> (s/explain boolean? false)
Success!
nil
cljs.user=> (require '[clojure.spec.gen.alpha :as gen])
nil
cljs.user=> (require '[clojure.test.check])
nil
cljs.user=> (gen/sample (s/gen boolean?))
(false true true false true true true true false false)

@jstaab cljs-test-runner should get you most of these things as well, I haven’t tried kaocha, it’s cljs support is pretty new

so you're saying, i could gen/sample some valid results to show the user? that's pretty cool

didn't think of that

@mfm I’m saying the spec you want is boolean?

@mfm Here you go: https://github.com/alexanderkiel/phrase

i agree that the spec that behaves correctly is boolean?; i was looking for a spec whose explanation says, "value foobar is not one of #{false true}"

why is that any better than not boolean?

for my intended audience, boolean is not obvious

thanks @didibus! will investigate

phrase looks great!

@borkdude If the audience is an end-user, boolean is bad feedback. You want the message to be simpler, and more contextualized to the app and the domain.

where exactly do you use this @mfm?

the intended audience is sort of an admin dashboard for our "tech support engineers" who have to diagnose mis-parsed client data. so it'd be great to have "value foobar is not one of #{some acceptable values}"

ok, cool

that is, my expectation is that they can look past some syntactic idiosyncracies ( #{} and the like ) but i'd still like it to be really really concrete

agreed that it's a relatively rare use case

you could also hook your own explain printer thing maybe

yeah i've been reading into that a little bit. and, in fact, using gen/sample is a great idea

you could also implement an explaining layer that maps a spec to a freeform string customized to your audience

{boolean? "should be either 'true' or 'false'" number? "should be be a machine readable number like 1 or 3.14"} etc.

you can get pretty far on specs almost doing the right thing with explain, but I'd assume you'd keep getting more awkward messages and corner cases

typically the expound + orchestra combo is good enough for me

I think that would depend on who is intended to interpret the message, if the reader isn't a dev (not to mention not a clojure dev), I'd expect a lot of awkward details

true

eg. even something like "string" can be confusing for a technical user who isn't a dev

i guess writing your own spec introspection util is one way to get this "right", so that you can optimize for your specific audience

in that case you could also write your own predicates and give them less technical names

the phrase library is another solution

that's true

great comments, thanks

@mfm Sorry if I missed this, but if you are in a position where you can define your own spec for boolean, you can use expound to set a custom error message

(require '[expound.alpha :as expound])
(require '[clojure.spec.alpha :as s])

(set! s/*explain-out* (expound/custom-printer {:print-specs? false}))

(s/def ::boolean boolean?)

(expound/defmsg ::boolean "should be either true or false")

(s/explain ::boolean false)
;; Success!
(s/explain ::boolean nil)
;; -- Spec failed --------------------

;;   nil

;; should be either true or false

;; -------------------------
;; Detected 1 error

You can’t associate a message with the boolean? predicate itself, but you can associate it with spec keywords that are predicates

@bbrinck Ah, now I get why you wanted those keyworded specs in speculative 🙂

indeed 🙂

very cool. have you used them anywhere yet?

I haven’t yet used them in speculative, but when I get some more free time, I have an idea about building a set of beginner-friendly messages for lots of core predicates

expound + custom messages + speculative + rebel-readline

awesome. I think I want a custom KLIPSE repl (faster performance) online with exactly that too

that would be sweet

Outside of the beginner context, I know that lucinia uses this functionality to clarify some complicated specs https://github.com/walmartlabs/lacinia/blob/d351d3c4dd9a0aad5cd078c87fabed0df2b14471/src/com/walmartlabs/lacinia/expound.clj#L22-L28

@bbrinck I’ve been wondering if expound can handle (s/or :f ::foo :b ::bar) well, no need to put composites in their own keyword right?

I suppose it depends on what you mean by “handle” 🙂

This is what happens by default:

(s/def ::foo int?)
(s/def ::bar string?)
(s/def ::foobar (s/or :f ::foo  :stuck_out_tongue: ::bar))

(s/explain ::foobar :abc)
;; -- Spec failed --------------------

;; :abc

;; should satisfy

;; int?

;; or

;; string?

;; -------------------------

@borkdude do you mean handle custom messages for :foo or :bar?

:D :stuck_out_tongue:

lol

I was going to edit, but that’s too good

@bbrinck yeah, so if you would have custom messages for either spec they would still be printed here?

Individual messages will work

(expound/defmsg ::foo "should be an integer")
(expound/defmsg ::bar "should be a string")
(s/explain ::foobar :abc)
;; -- Spec failed --------------------

;;   :abc

;; should be an integer

;; or

;; should be a string

;; -------------------------

but if you want a message for the whole thing you have to wrap in a predicate

https://github.com/bhb/expound/issues/101

it would also be helpful to have a 4clojure instance running with expound + speculative I bet

@bbrinck the or printing is good, thanks!

Hello - I've tried uploading a simple project to Clojars (cljblpapiwrapper). That project uses a local jar file (non Clojure) as a dependency, declared in :resource-paths in project.clj. Now upon trying to use this project somewhere else, it fails to locate the jar file. How should I go around the problem? Can I publish the jar on Clojars separately? Thank you,

the natural way to share deps is to make maven artifacts, and it's straightforward to use clojars for that, sure

or they could check out the repo and run lein install

If a clojure file defines a private function like (defn- a-func [] true) is there a way to call it like (in-ns 'my-private.ns) (my-private.ns/a-func)?

it is a bad idea to put in-ns forms in the middle of a source file unless you really know what you are doing

Sorry I mean call it from within a repl

@#'a-func will let you call the private function

(which manually does the bits the compiler would do for you if it wasn't so worried about things being private)

Hmm shouldn’t that work? https://github.com/weavejester/environ/blob/master/environ/src/environ/core.cljc#L80 I’m trying to run the following in a REPL

(require 'environ.core)
(in-ns 'environ.core)
(def env (#'read-env))

don't do the in-ns

make sure the version of the code you are reading matches the version of the mode your are running

Oh, good point

😵 That was the issue. It was refactored since the latest release. Thanks for helping me catch that.